View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by PubMed Central The Carboxyl Terminus Coaligns with and Specifically Disrupts Intermediate Filament Networks When Expressed in Cultured Cells Thaddeus S. Stappenbeck and Kathleen J. Green Department of Pathology and the Cancer Center, Northwestern University Medical School, Chicago, Illinois 60611

Abstract. Specific interactions between tides including the 90-kD carboxy-terminal globular I and 11 (DP I and II) and other desmosomal or cyto- domain of DP I specifically colocalized with and ulti- skeletal molecules have been difficult to determine in mately resulted in the complete disruption of IF in part because of the complexity and insolubility of the both lines. This effect was specific for IF as micro- and its constituents . We have used a mo- tubule and networks were unaltered . lecular genetic approach to investigate the role that This effect was also specific for the carboxyl terminus DP I and 11 may play in the association of the desmo- of DP, as the expression of the 95-kD rod domain of somal plaque with cytoplasmic intermediate filaments DP I did not visibly alter IF networks. Immunogold (IF) . A series of mammalian expression vectors en- localization of COS-7 cells transfected with constructs coding specific predicted domains of DP I were tran- including the carboxyl terminus of DP demonstrated siently expressed in cultured cells that form (COS-7) an accumulation of mutant in perinuclear aggre- and do not form (NIH-3T3) . Sequence gates within which IF subunits were sequestered. encoding a small antigenic peptide was added to the 3' These results suggest a role for the DP carboxyl ter- end of each mutant DP cDNA to facilitate immuno- minus in the attachment of IF to the desmosome in ei- localization of mutant DP protein. Light and electron ther a direct or indirect manner. microscopical observations revealed that DP polypep-

DMOSOMEs are intercellular junctions that function Ultrastructurally, desmosomes appear as disc-shaped in cell- and act as specific cell surface regions, 0.5-2.0 Am in diameter and -100-run thick, sym- attachment sites for intermediate filaments (IF), metrically arranged about two adjacent cell membranes. Be- (for reviews see Steinberg et al., 1987; Garrod et al., 1990; tween the plasma membranes is a 20-30-nn space which Green andJones, 1990; Schwarz et al., 1990). The supracel- contains the extracellular domains of the transmembrane lular network of cytoplasmic IF and desmosomes is found glycoproteins of the desmosome (or desmoglea). The cyto- in a number of tissues such as epithelia, myocardium, and plasmic portion of the desmosome consists of an electron- arachnoid where desmosomes interact with , , dense, trilaminar plaque of variable thickness (10-40 nn), and IF, respectively (Schwarz et al., 1990). While which underlies both cell membranes (Schwarz et al., 1990). the exact function of IF has yet to be fully understood IF are thought to interact with the desmosome by looping (Steinert and Roop, 1988 ; Klymkowsky et al., 1989), the ul- through the innermost plaque region (Kelly, 1966). trastructure of the IF-desmosome network suggests that in The highly related molecules, desmoplakins I and II (DP addition to maintaining the integrity and proper relationship I and II), are the most abundant constituents ofthe cytoplas- among cells, it may impart increased tensile strength and me- mic plaque region of the desmosome (Mueller and Franke, chanical resistance to whole tissues (Arnn and Staehelin, 1983; Kapprell et al ., 1985). Based on an analysis of the 1981). Recent work that supports this idea demonstrated that predicted amino acid sequence (Green et al., 1990), DP I is transgenic mice with disrupted IF in the basal layer of strat- predicted to form a homodimer comprising a central «-he- ified epithelia displayed abnormalities in epidermal architec- lical coiled-coil dimer -130-nn long, flanked by two globu- ture and blistered easily (Vasser et al., 1991). Although cy- lar ends corresponding to the amino and carboxy termini of toplasmic IF have been extensively investigated, the exact themolecule. This prediction is consistent with the biochem- nature and importance of the link to desmosomes is un- ical evidence that purified and crosslinked DP I exists as a known. dimer in vitro. Furthermore, rotary shadowed images of 1. Abbreviations used in this paper: DP, desmoplakins ; IF, intermediate fila- purified DP I (OKeefe et al., 1989) appeared dumbbell ments; PCR, polymerase chain reaction. shaped with a central "rod" of the length predicted by Green

© The Rockefeller University Press, 0021-9525/92/03/1197/13 $2 .00 The Journal of , Volume 116, Number5, March 1992 1197-1209 1197 et al. (1990). The rod domain of DP I is characteristic of Construction ofExpression Vectors many a-fibrous including IF in that it is predomi- For each construct, polymerase chain reaction (PCR) mutagenesis was per- nantly composed of a series of heptad repeats (Conway and formed to generate sequence for a molecular tag, an artificial start, and a Parry, 1990). DP II is thought to be derived from an alterna- stop site. AmpliIàq DNApolymerase (Perkin-Elmer Corp., Norwalk, CT) tively spliced mRNA of DP I resulting in a greatly shortened was used according to the manufacturer's instructions in a reaction with 10 rod domain (Green et al., 1990). ng of the appropriate DP I cDNA and 1 gM of the appropriate oligos. After PCR amplification for 35 cycles (1 min at 94°C, 2 min at 45°C for cycles One particularly interesting feature of the 851 amino acid 1-5, and 2 min at 55°C for cycles 6-35, 1 min at 72°C), products were iso- carboxyl terminus of DP I and II is a series of three regions lated from a polyacrylamide gel, subcloned into pGEM-9Zf(-) (Promega that contain 4.6 copies of a 38-residue repeat (Green et al., Corp., Madison, WI) and sequenced using the Sequenase Kit (United States 1990). The periodicity of the acidic and basic residues of Biochemical, Cleveland, OH) to confirm that the correct sequences had been inserted . these repeats matches that of the 1B rod domain of IF thus Construction of pDPCT. A 34-mer oligonucleotide 5'-AAGAGCT indicating a basis for potential ionic interaction between the CGCCATGGgagcatctgcttctccta-3' (Northwestern Biotechnology Research two molecules . The possibility that DP links IF to the des- Service Facilities, Chicago, IL) was used to insert sequence encoding a Sacl mosome has been previously suggested based on a number restriction site followed by a portion of the Kozak (Kozak, 1986) consensus sequence for binding and an ATG start site (both underlined) 5' of observations (Green and Jones, 1990). Furthermore, the of the predicted carboxyl terminus (lower case). This oligo in the sense 38-residue repeat has also been found in the carboxyl termi- orientation was used in a PCR reaction with an antisense oligo 5'-CTGT nus of , which is a known IF-associated protein CGACAGTCAGCTT 3' that was located at an internal Sall site 150-bp (Wiche et al., 1991), and the 230-kD bullous pemphigoid an- downstream from the start site of the carboxyl terminus . This 150-bp tigen (Green et al., 1990; Tanaka et al., 1991), which has SacI/Sall fragment was subcloned into a plasmid containing a 2 .4-kb SaII/EcoRI fragment ofDP cDNA encoding a majorportion ofthe carboxyl been localized to the plaque region of . terminus (pl) . A 71-mer oligo 5'-agcagtagttctattgggcac GTGGAGCAAAAG- Hemidesmosomes act as attachment sites for IF at the CTCATTTCTGAAGAGGACTTGTAGGGTACCGAATTCCC-3' was used dermal-epidermal junction (Green and Jones, 1990), so it is to insert sequence encoding a 33-bp c-myc epitope (underlined), a stop intriguing that this junction contains a plaque component codon and a KpnI and an EcoRl restriction site at the 3' end of the DP I similar to DP in desmosomes. cDNA (lower case). This oligo in the antisense orientation was used in a PCR reaction with a sense oligo 5'-GTAGGAAGAATTCCTGC-3' that was One difficulty with the proposed DRmediated linkage of located 70-bp upstream from the end of DP I at an EcoRl site . The EcoRI IF to the desmosome is that previous attempts using bio- fragment was subcloned into pl. The full DPCT was then cut out with a chemical techniques have been unable to demonstrate an in- SacI/KpnI digest and blunt end ligated into pRC4B (kindly provided by Dr. teraction between DP and IF or any other molecule (O Keefe R. Scarpulla, Northwestern University, Evanston, IL) . Construction of pDPROD. A 35-mer oligonucleotide, 5'-AAGAGCT et al ., 1989 ; Pasdar et al ., 1991) . There are several possible CGCCATGGagaaagccatcaaggagaag-3' was used to insert sequence encoding explanations for these results. First, ifDP does directly bind a SacI restriction site followed by aportion of the Kozak consensus sequence IF, it is possible that the denaturing conditions used to ex- for ribosome binding and an ATG start site (both underlined) 5' of the tract DP from desmosomes during purification affected the predicted rod domain (lower case). A 49-mer oligonucleotide, 5'-TTGAGC- in binding capacity of DP Another possibility is that TCCTACATGAGGCCGAAGAAagcgatagatcctgcaccccgaa-3' was used to in- vitro sert sequence encoding a 15-bp fragment of the neuropeptide substance P DP may interact directly with IF, but only in the presence (underlined), a stop codon and a SacI restriction site 3' of the predicted end of accessory proteins that would stabilize the interaction. A of the rod domain (lower case). These twooligos (the former sense, the lat- third possibility is that one or more linking proteins mediate ter antisense) were used in a PCR reaction to generate a 2.7-kb Sacl frag- an indirect association of DP and IF. The possibilities that ment which was subcloned into pGEM9Zf(-) . An internal 2.4-kb BglII/EcoRI fragment was removed and replaced with the same fragment accessory or linking proteins may be involved must be con- from DP I cDNA . The flanking PCR-generated portions of the construct sidered as the desmosomal plaque is complex and a number were sequenced to ensure that there were no errors in this region because of other desmosomal molecules have been proposed as IF of PCR. The Sacl fragment was then subcloned into pRC4B. linkers (Tsukita and Tsukita, 1985 ; Kapprell et al., 1988; Construction pDPAN. A Sacl/BglII fragment from pDRROD which ., ; Foisner et al., 1991). To avoid difficul- encoded the start translation site was ligated into a plasmid containing a Cartaud et al 1990 BgIII/EcoRI fragment of DP I cDNA . An EcoRI/KpnI fragment from ties inherent with in vitro binding experiments, we have cho- pDPCT which encoded the tag and stop codon was subcloned into the sen a molecular genetic approach to test the possible interac- previous plasmid . A Sacl/KpnI fragment was then subcloned into pRC4B. tion between the carboxyl terminus ofDP and IF in cells that AdditionalConstructs. Expressionplasmids were constructedwith both would contain any putative cofactors or linking proteins. tags using a similar approach . In addition, tagless constructs for both pDP.CT and pDPAN were generated using PCR primers not containing the Using the previously characterized partial cDNA of DP I, tag sequences. the carboxyl terminus with and without the rod domain of DP I was expressed in tissue culture cells. By tracking the DNA Transfections pattern of expression of mutant protein by immunofluores- carboxyl ter- DNA was transfected into COS-7 and NIH-373 cells using the calcium cence and immunogold EM, we found that the phosphate method (Graham and Van der Eb, 1973) followed by a 15% minus of DP specifically coaligned with and eventually glycerol shock (Parker and Stark, 1979). Cells were fixed 48 h after glyc- resulted in the disruption of IF networks. erol shock unless otherwise noted in the text.

Preparation and Immunoblotting of Materials and Methods Whole Cell Extracts Culture Whole cell extracts were prepared as described by Green et al . (1991) . Pro- Cell tein content was determined by the method of Bradford (1976) . Samples of COS-7 African Greenmonkey kidney cells that constitutively express SV-40 30 lAg were loaded and run on SDS-polyacrylamide gels (6.5 %). The gels large T antigen, thus replicating plasmids with an SV-40 origin of replica- were transferred to nitrocellulose and immunoblotting was performed as tion to a high number (Gluzman, 1981) and NIH-3T3 mouse previously described (Angst et al., 1990). Primary antibodies used were a were cultured in DME plus 10% FCS and 100 U/ml penicillin and 100 1:2,000 dilution of an affinity-purified rabbit polyclonal, NW 6, directed l+g/ml streptomycin . against a fusion protein of the carboxyl terminus of DP (Angst etal., 1990);

The Journal of Cell Biology, Volume 116, 1992 119 8 a 1 :500 dilution of a mouse monoclonal, DPI-2 .17, directed against DP I (Cowin et al ., 1985) ; and a 1 :10 dilution of a mouse monoclonal, 9E10.2, directed against an 11 amino acid fragment of human c-myc (Evans et al ., 1985) . Primary antibody binding was detected with a 1 :1,000 dilution of peroxidase-coupled anti-rabbit or anti-mouse secondary antibodies (Kirke- gaard and Perry Laboratories, Inc., Gaithersburg, MD) .

Immunofluorescence Labeling For immunohistochemistry, cells were grown on glass coverslips . After transfection, cells were washed in PBS and fixed in either methanol (-20°C) for 2 min or 4% paraformaldehyde (25°C) for 10 min . Cells fixed with paraformaldehyde were extensively washed in PBS and were permea- bilized by incubation in 0.2% Triton X-100 in PBS on ice for 5 min . Primary antibodies used for staining were a 1:10 dilution of a mouse monoclonal, 9E10.2 directed against an 11 amino acid fragment of human c-nryc (Evans et al ., 1985) ; a 1 :50 dilution of a rat monoclonal antibody, NCI/34, directed against substance P (Accurate Chem . & Sci. Corp., West- bury, NY) ; a 1 :50 dilution of an affinity-purified rabbit polyclonal antise- rum, NW 6, directed against a fusion protein of the carboxyl terminus of DP (Angst et al ., 1990) ; a 1 :20 dilution of a mouse mAb directed against DP I and II (Boehringer Mannheim Corp ., Indianapolis, IN) ; a 1 :10 dilution Figure 1. Expression vectors encoding the predicted carboxyl ter- of a rat mAb, Tromal, directed against mouse K8 (kindly provided by Dr. minus (pRC4DPCT), the predicted rod domain (pRC4DPROD), K . Trevor and later obtained from the Developmental Studies Hybridoma and the carboxyl terminus plus the rod domain (pRC4DPAN) of Bank under control NOI-HD-6-2915 from the National Institute of Child DP I . These constructs are referred to in the text as pDPCT, Health and Human Development) ; a 1 :5 dilution of a mouse mAb, RGE 53, pDRROD, and pDRON, respectively . directed against human K18 (ICN Immunobiologicals, Lisle, IL) ; a 1:50 di- lution of a rabbit polyclonal anti-vimentin antiserum (ICN Immunobiologi- cals) ; and a 1 :10 dilution of a mouse monoclonal, E7, directed against 0- (Developmental Studies Hybridoma Bank) . A 1 :10 dilution of (amino-terminal deletion designated DRAN) (Green et al ., rhodamine phalloidin (Molecular Probes, Eugene, OR) was used as a probe 1990) were cloned into the eukaryotic expression vector, for f- . All antibodies were diluted in PBS . pRC4B (Evans and Scarpulla, 1988) . This vector uses the rat To visualize the primary antibody, appropriate fluorescein- or rhoda- cytochrome c promoter and intron enhancer sequences to mine-conjugated anti-rabbit, anti-mouse (Kirkegaard and Perry Laborato- ries), or anti-rat (Jackson ImmunoResearch Laboratories, Inc., West Grove, drive the expression of a cDNA insert . In addition, it en- PA) secondary antibodies were diluted 1 :20 in PBS . Controls included incu- codes an SV-40 origin of replication, which increases the bation of fixed cells in conjugated secondary antibodies alone and with copy number of the plasmid in cells such as COS-7 that ex- preimmune rabbit serum for polyclonal antibodies . press large T antigen (Gluzman, 1981), thus further enhanc- ing expression . Domain-specific constructs of DP cDNA EM: Conventional and Immunogold Labeling were generated by a combination of PCR mutagenesis and Transfected COS-7 cells were processed for conventional EM as previously routine subcloning techniques . DNA sequence analysis was described (Green et al ., 1991) . For immunogold labeling, cells were grown used to verify structure of the completed constructs (for de- and transfected on nonetching plastic Permanox tissue culture dishes (Elec- tails see Materials and Methods) . Peptide tags encoding ei- tron Microscopy Sciences, Fort Washington, PA) . The cells were fixed over- night in 4% paraformaldehyde plus 0.1% glutaraldehyde in 0.1 M sodium ther a 5-amino acid fragment of substance P (Albers and phosphate (pH 7.4) . The cells were then washed in cacodylate buffer and Fuchs, 1987, 1989) or an 11-amino acid epitope of c-myc dehydrated in a 30, 50, and 70% ethanol series . This was followed by two (Munro and Pelham, 1987) were used to follow the expres- washes of a mixture of 70% ethanol and LR White resin (Electron Micros- sion ofmutant DP in immunolocalization experiments . Pep- copy Sciences) for 1 h each . The cells were infiltrated in LR White for 24 h tide tags were used in part because species-specific DP anti- at room temperature . Fresh LR White was then added and polymerized at 60°C in the presence of nitrogen gas . Sections were placed on nickel grids bodies were not readily available. Two different tags were which were used for immunogold labeling . used to ensure that observations concerning the effects ofeach Grids were blocked in a 1 :20 dilution of normal goat serum in TBS plus domain were not affected by a specific peptide tag . For sim- I% BSA (TBS-BSA) for 30 min (all steps at room temperature) . Primary plicity, the constructs in Fig . 1, pRC4DPCT, pRC4DP ROD, antibodies used for staining were a 1 :20 dilution of a mouse monoclonal, 9E10.2, and a 1 :200 dilution of a mouse monoclonal, V9, directed against and pRC4DPON, will be referred to herein as pDPCT, vimentin (ICN ImmunoBiologicals) . These antibodies were diluted in pDPROD, and pDPAN, respectively. TBS-BSA and applied to grids for 1 h . The grids were then washed in These constructs were transiently transfected into COS-7 TBS-BSA plus 0 .1% Tween and then were incubated for 1 h with a 15-nm cells by calcium phosphate precipitation and cell extracts gold-labeled goat anti-mouse antibody (Amersham Corp., Arlington were processed 48 h after glycerol shock for analysis by Heights, IL) that was diluted 1 :10 in TBS-BSA . The grids were then washed again in TBS-BSA plus 0.1% Tween . In specific experiments noted in the SDS-PAGE . A rabbit polyclonal antibody directed against a text, grids were counterstained with 3 % uranyl acetate for 1 min and lead fusion protein of the carboxyl terminus of DP used for im- citrate for 1 min . Controls included incubation without a primary antibody munoblotting (Angst et al ., 1990) reacted only with the 90 and with the mouse mAb, E7, against ß tubulin . and 185-kD proteins expressed by pDPCT and pDP ON, respectively, as well as the endogenous 240 and 210-kD DP I and II . (These values reflect relative migration on SDS- Results PAGE . The actual molecular masses ofDP I and II based on the predicted amino acid sequence are 310 and 238 kD, Expression ofMutant DPDomains in COS-7 Cells respectively ; Virata et al . 1992) (Fig. 2) . A DP I-specific cDNA sequences precisely corresponding to the predicted mAb, DPI-2 .17 (Cowin et al ., 1985), which has been epitope carboxy-terminal globular domain (DRCT), the central rod mapped to the DP I-specific region of the rod (Nilles, L ., domain (DP ROD), and the carboxyl terminus plus the rod and K . Green, unpublished results), reacted with only the 95

Stappenbeck and Green DP COOH Terminus-mediated Collapse of IF 1199 Figure 2 . Detection of mutant DP ex- pressed in COS-7 cells by immunoblot- ting. COS-7 cultures were transfected with each DP construct and were lysed in urea sample buffer at 48 h after glycerol shock . 30 pg of total protein from COS-7 transfected with no DNA (lanes 1), pDP.ROD (lanes 2), pDPCT (lanes 3), or pDPON (lanes 4) were loaded and run on a 6.5 % SDS-polyacrylamide gel . These gels were transferred to nitrocellulose and incubated with a rabbit polyclonal anti- body directed against the carboxyl termi- nus of DP (NW 6), a mouse monoclonal directed against DP I (DPI-2.17) or a mouse monoclonal directed against an 11- amino acid epitope of c-myc (9II0.2) . The latter antibody recognizes only lanes 3 and 4 as pDPROD is tagged with sub- stance P in this experiment . Arrows indi- cate molecular mass standards of 200, 116.5, 97.4, 66.2, and 42 .7 kD, from top to bottom . and 185-kD proteins expressed by pDPROD and pDPON, nous DP (Fig . 2) . To make a quantitative estimate on a popu- respectively, as well as the endogenous DP I . The 9E10.2 lation basis, transfected cells were metabolically labeled and mAb directed against an 11-amino acid residue fragment of extracts were immunoprecipitated using a polyclonal anti- c-ntyc reacted on immunoblots only with protein expressed body to DP I that was reactive to all three mutants (not by pDPCT and pDPON tagged with this epitope (Evans et shown) . taking into account transfection efficiency, the rela- al ., 1985). Antibodies specific for a fragment of substance tive amount of protein expressed by each construct was ap- P reacted only with proteins tagged with this epitope (not proximately equivalent and was 15-20-fold higher than en- shown) . dogenous DP I as determined by densitometric scanning of From the immunoblots, the relative amounts of each DP autoradiograms . domain expressed appeared to be approximately equal and Immunolocalization of protein expressed from each con- in each case appeared to be many times higher than endoge- struct 48 h after glycerol shock demonstrated interesting and

Figure 3. Indirect double label immunofluorescence of COS-7 cells transfected with pDP.CT (A and B) or pDPON (C and D) 48 h after glycerol shock . A and C were stained with the rabbit polyclonal anti- body, NW 6. B and D were stained with the mouse monoclonal, 9E10.2, which is specific for the c-ntyc tag on transfected mutant DP. Long arrows in all panels identify the perinuclear aggregate . Because of the intensity of this aggregate in C and D, two dif- ferent exposures are shown for each panel to clearly demonstrate the cytoplasmic spots (short arrows) . Bar, 10 pm .

The Journal of Cell Biology, Volume 116, 1992 1200 distinct phenotypes. Immunofluorescence analysis using an- differences in the DPCT and vimentin patterns were ob- tibodies to DP or to the c-myc tag revealed the presence of served, presumably because of the additional colocalization large perinuclear aggregates as well as diffuse cytoplasmic ofmutant protein with the keratin network. These data were staining in cells expressing pDPCT (Fig. 3, a and b). The obtained using the 9E10.2 antibody to the c-myc tag (similar expression ofpDP.AN also resulted in the formation of large results were obtained with DP antibody), thus confirming perinuclear aggregates as well as an array of discrete spots that the colocalization was with exogenous mutant DP At in the (Fig. 3, c and d) . These cytoplasmic spots times 9-24 h after glycerol shock, marry transfected cells appeared to be similar to those formed by the protein ex- contained bundles of IF that colocalized with mutant DP ex- pressed by pDP ROD in transfected cells (see Fig. 5a). pressed from either pDPCT or pDP ON. These IF appeared to be in the process ofdisruption as by 24 h most were com- Effect ofMutant DP on Cytoskeletal Systems of pletely disrupted in transfected cells. The result of the time COS-7 Cells course experiment indicated that the IF disruption was not of double-label immunofluorescence experiments because ofthe formation ofa mutant protein aggregate which A series IF . Similar colocaliza- were performed to determine if endogenous cytoskeletal sys- then secondarily caused to collapse tems were affected by the overexpression of these mutant tion was also observed when the amount of plasmid used in transfection decreased by two-thirds and cells were proteins. We were particularly interested in possible effects the was observed COS-7 cells express both assayed at 24 h after glycerol shock. The colocalization and on IF networks. We that with DP carboxyl termi- type III vimentin and type I and II , although the ex- the disruption ofIF were observed pression keratin filaments was decreased in sparse cultures nus-containing constructs that were tagless or tagged with of P typically used for transfection. This phenomenon has been either substance or c-myc. previously observed with cultured epithelial cells (Rhein- COS-7 cells transfected with pDPCT and pDP ON were also assayed at 72-144 h after glycerol shock. At these later wald and O'Connell, 1985) . We also observed that COS-7 synthesizing desmosomal components, time points, colonies of transfected COS-7 cells with dis- cells were capable of of trans- which is consistent with the epithelial nature of these cells. rupted IF were observed. By 144 h, the percentage under the culture conditions used for transient fected cells as determined by immunofluorescence was However, greatly continuous overexpres- transfection, desmosomes were formed only occasionally. decreased suggesting that the Therefore we focused our analysis on the effects of each DP sion of these constructs was ultimately toxic to cells. Im- remaining viable cells expressing domain on IF. munofluorescence of the indirect immunofluorescence of COS-7 cells mutant DP typically demonstrated either the presence of Double label or colocalization transfected with either pDPCT or pDPON revealed a strik- short IF around the perinuclear aggregate ing reorganization of both keratin and vimentin networks of mutant DP with completely reformed IF networks. It is into tight, often perinuclear, aggregates. Fig. 4, b and d possible that these cells were in the process of recovery and demonstrates this for pDPCT and similar results were ob- expressed very little, if any, mutant protein. tained with pDPAN (not shown). In both cases, the dis- rupted IF colocalized with the perinuclear aggregate of mu- Examination of Transfected COS-7 Celts by EM tant DP At this level of expression, >90% of transfected The aggregates of mutant protein and disrupted IF in COS-7 cells had completely disrupted IF, as compared to untrans- cells transfected with pDPCT and pDPON were analyzed in fected COS-7 cells in which 10-15% of the cells were typi- more detail at the ultrastructural level by both conventional cally observed to have disorganized IF networks. This pat- and immunogold EM. The perinuclear aggregates observed tern in untransfected cells is possibly a result ofthe transient at the light microscope level in cells transfected with pDPCT reorganization of IF shown to occur in certain cells during appeared to comprise an electron-dense accumulation of mitosis (Aubin et al., 1980) . The disruption was specific for smaller aggregates in the cytoplasm of the cell that were nei- IF as neither the nor the microfilament systems ther located within vesicles nor associated with the nuclear were affected (Fig. 4, e-h) . Intriguingly, the perinuclear ag- membrane (Fig. 7, a and b) . Immunogold labeling of trans- gregate colocalized with the microtubule organizing center fected cells demonstrated that the aggregates contained both (Fig. 4, e and f) . IF disruption was also specific for con- mutant DP and vimentin throughout the entire structure structs containing the carboxyl terminus of DP, as cells that (Fig. 7, c and d), implying that the IF networkdid not simply expressed the rod domain of DP I alone did not have any ap- collapse around a preformed aggregate of mutant DP As a parent disruption of vimentin (Fig. 5, a and b) or keratin control, a mAb to ß-tubulin specifically labeled the cyto- (Fig. 5, c and d) networks. As in untransfected cells, 10-15 % plasm, but not the aggregates in transfected cells (Fig. 7 e). of cells transfected with pDPROD were observed to have Mutant protein in cells transfected with pDPON localized disorganized vimentin networks that did not colocalize with to two distinct types of aggregate (Fig. 8, a and b). One type the cytoplasmic aggregates of protein expressed from this of aggregate was similar to the perinuclear aggregates of vector. DPCT in that it contained both mutant DP protein and The nature of the carboxy-terminal-induced IF disruption vimentin throughout (Fig. 8, c and d) . However, the ultra- was investigated by examining earlier times after transfec- structure ofthe vimentin-containing DP AN aggregates was tion, using the same amount of the expression vector DNA. quite different. While aggregates of DPCT were electron Mutant DP was first detected by immunofluorescence 9 h af- dense with no apparent filamentous substructure, aggregates ter glycerol shock. Protein expressed from both pDPCT and of DR AN were composed of a filamentous meshwork. The pDPAN displayed a striking colocalization with intact IF fine filaments of this meshwork were shorter and of smaller (Fig. 6, a-d), while the protein expressed from pDP ROD diameter than the 8-10-run cytoplasmic IF found in neigh- was diffuse in the cytoplasm (not shown) . Occasional local boring nontransfected cells. The second type of aggregate in

Stappenbeck and Green DP COON Terminus-mediated Collapse of IF 120 1

The Journal of Cell Biology, Volume 116, 1992 1202

Figure S. Indirect double label immunofluorescence ofCOS-7 cells transfected with pDPROD tagged with substance P, 48 h after glycerol shock . a and b were stained with DPI-2 .17 and a vimentin polyclonal antibody, respectively. c and d were stained with a rabbit polyclonal antibody directed against DP and RGE 53, directed against K18. Bar, 10 Am .

cells transfected with pDPAN was electron dense, nonfila- ence ofother desmosomal proteins present in these cells that mentous, and did not contain vimentin (Fig. 8, e and f) . acted as cofactors or linking proteins. To test this, mutant DP Their appearance was similar but not identical to the cyto- constructs were transiently transfected into NIH-3T3 mouse plasmic aggregates observed in cells transfected with fibroblasts that do not form desmosomes . In each construct, pDPROD, which also were not labeled by a vimentin mAb the patterns of expression observed by double-label indirect (not shown) . It should also be noted that an unrelated anti- immunofluorescence were similar to those seen in COS-7 body directed against ß-tubulin did not label either type of cells . Immunofluorescence analysis of NIH-3T3 cells trans- aggregate, but was restricted primarily to the cytoplasm . fected with either pDPCT or pDRAN 48 h after glycerol shock, revealed similar patterns of expression observed at various time points in COS-7 cells . At 48 h after glycerol Expression ofMutant DP Domains in NIH-3T3 Cells shock, mutant protein colocalized with intact vimentin net- Resulted In a Similar Pattern ofImmunolocalization works in certain cells (Fig . 9, a and b) and disrupted vimen- It is possible that the apparent association ofthe carboxyl ter- tin networks in others (Fig. 9, c and d) . This variable pheno- minus of DP with IF in COS-7 cells was because of the pres- type was likely to be a result of different levels of expression

Figure 4. Indirect double label immunofluorescence of COS-7 cells transfected with pDPCT (c-myc tagged in a, b, g, and h, and substance P tagged in c-f) 48 h after glycerol shock . a and b were stained with 900 .2 and a rabbit polyclonal antibody directed against vimentin, respectively . c and d were stained with NW 6 and a mouse mAb directed against K18 (RGE 53), respectively . Both the vimentin and keratin networks were collapsed in transfected cells and colocalized with the perinuclear aggregate of mutant DP e andfwere stained with NW 6 and a mouse mAb directed against ß tubulin (E7) . g and h were stained with 900.2 and rhodamine phalloidin to label actin filaments . Neither the microtubule nor the microfilament systems were altered in the transfected cells . The results were identical using the c-myc or the substance P tag . Arrows indicate the location of the perinuclear aggregate . Bars, 10 ym .

Stappenbeck and Green DP COON Terminus-mediated Collapse of IF 1203 Figure 6. Indirect double label immunofluorescence of COS-7 cells transfected with DDPCT (a and b) or pDPON (c and d) 9 h after glycerol shock . a and c were stained with 9E10.2, to identify c-myc tagged mutant DR b and d were stained with a vimentin polyclonal antibody. At this time point, both mutant DPs colocalize with the vimentin IF system in transfected cells . Bar, 10 jAm . among transfected cells . Both the colocalization and disrup- cells using a vector with a strong promoter and an SV-40 ori- tion were noted to be more pronounced with DR ON, possi- gin ofreplication (Blouin et al ., 1990) . While K18 incorpo- bly because of its potential to dimerize . This result implies rated into the endogenous filament network, the "excess K18 that proteins restricted to desmosomes may notbe absolutely accumulated in a juxtanuclear site similar to the aggregates required to either stabilize or mediate the observed associa- produced in cells overexpressing DPCT. However, in cells tion between carboxy-terminal containing mutants of DP overexpressing K18, the keratin network remained un- and IF. disturbed and the vimentin network was only partially reor- ganized . Therefore overexpression ofan insoluble cytoskele- tal molecule is not sufficient to Discussion disrupt IF even when the molecule can interact normally with endogenous IF at lower We have presented evidence that truncated DP comprising levels of expression . the predicted carboxyl terminus specifically colocalizes with In addition to the disruption of IF observed in cells that IF in COS-7 and NIH-3T3 cells. When expressed athigh lev- overexpressed DPCT and DRON, we also observed coalign- els, IF networks in these cells are ultimately disrupted . Sev- ment without disruption in cells that were most likely ex- eral lines of evidence suggest that this effect is specific both pressing lower amounts of these mutant proteins . This evi- for the carboxyl terminus ofDP and for the IF system . First, dence argues for the specificity of the interaction. That the the rod domain of DP expressed in either cell line neither effects described above are specific for the IF system is also colocalized with nor disrupted IF. The possibility that differ- reflected in the nature ofthe disruption andthe ultrastructure ing levels of expression were responsible for the apparent of the resulting aggregates . Electron microscopical analysis specificity was ruled out by quantifying levels of mutant DP revealed that at high levels ofexpression, IF did not just col- expression in transfected cells . Second, microtubule and lapse into whorls around the nucleus, as occurs in response microfilament systems were not obviously affected by the to heat shock (Shyy et al., 1989) or pharmacological agents overexpression of any domain ofDR Finally, other investiga- such as colchicine (for review see Klymkowsky et al ., 1989) . tors have overexpressed the entire cDNA in COS4 Instead, IF in transfected cells appeared to disassemble and

The Journal of Cell Biology, Volume 116, 1992 1204 Figure 7. EM of COS-7 cells transfected with pDPCT 48 h after glycerol shock . Cells were fixed either for conventional EM (a and b) or for immunogold EM (c-e) . Clusters of extremely electron-dense aggregates in cells transfected with this construct, at low (a) and high (b) magnification . Postembedment immunogold labeling of these aggregates was done with the mouse mAbs : 9E10.2, directed against the c-rayc peptide tag (c) ; V9, directed against vimentin (d) ; and E7, directed against 0 tubulin (e) . The gold particles conjugated to anti-mouse secondary antibodies were 15 run in diameter. Due to the density of these aggregates, the sections were not counterstained . Therefore, the aggregates appear to be less electron dense. The edge of labeled aggregates was shown in the figures to demonstrate the specificity for either the aggregate (c and d) or for the cytoplasm (e) . Note that under the conditions of LR White embedment without counterstaining cannot be visualized even though their antigenicity is retained (e) . N, nucleus. Bars : (a) 1 pm ; (b-e) 0.25 gym . subsequently, IF subunits became reorganized into electron Those formed with DPCT were electron dense and not dense (DPCT) or fine filamentous meshworks (DPAN) also noticeably filamentous while those formed with DPAN containing mutant protein . The apparent disassembly we ob- comprised a fine filamentous meshwork . In addition, non- served is more in line with observations by Klymkowsky vimentin-containing aggregates of DPAN or DPROD did (Klymkowsky, 1981 ; Klymkowsky et al ., 1983), in which not consist of fine filaments . It follows that to form filamen- microinjection of antibodies directed against intermediate tous aggregates, the rod domain and carboxy-terminal do- filaments resulted in a disappearance of the IF system . IF mains of DP as well as vimentin were all necessary . It is pos- disruption has also been observed as a result of express- sible that the rod domain of DP AN allowed the formation ing of dominant negative mutations of keratins (Albers and of dimers and/or oligomers that interact with vimentin to Fuchs, 1987,1989 ; Kulesh et al ., 1989 ; Lu and Lane, 1990 ; create higher order structures whose ultrastructure differed Trevor, 1990) and the 19-kD product of the ElB oncogene so dramatically from the dense aggregates formed by DPCT (White and Cipriani, 1989, 1990) . However, in these latter and vimentin. It may be noteworthy that the DPAN-vimentin- cases, an ultrastructural analysis of the affected IF networks containing aggregates ultrastructurally resemble the mesh- was not performed . Therefore, we cannot speculate on the work of 4-5-run fine filaments that comprise the innermost potential similarities with the disrupted filaments we ob- desmosomal plaque (Pirbazari and Kelly, 1985) . It is possi- served . ble that by overexpression of this portion of DP which can As described above both DPCT and DPAN were capable associate with itself and vimentin, a large plaque-like struc- of forming aggregates with vimentin . However, as demon- ture is assembled in the cytoplasm . Expression of full-length strated by conventional and immunogold EM, the vimentin- DP including the amino terminus may facilitate the localiza- containing aggregates were ultrastructurally quite different . tion of this meshwork to the cell cortex .

Stappenbeck and Green DP COON Terminus-mediated Collapse of IF 1205

Figure 8. EM of COS-7 cells transfected with pDPON 48 h after glycerol shock . Cells were fixed either for conventional EM (a and b) or for immunogold EM (cf) . Large filamentous aggregates were observed near the nucleus (N), as shown in a, and many smaller, electron- dense spots were seen in the cytoplasm of cells transfected with this construct (as shown in b) . Postembedment immunogold labeling of transfected cells was carried out with the mouse mAbs: 9E10.2, directed against the c-myc peptide tag (c and e) and, V'9, directed against vimentin (d andf) . c and d are high magnifications of the filamentous perinuclear aggregate that were vimentin positive . e andfare high magnifications of the cytoplasmic electron-dense spots that were vimentin negative. These sections were counterstained so that the fine filamentous meshwork was visible . Bars : (a) 5 Am; (b) 1 Am ; and (cf) 0.25 Am .

The Journal of Cell Biology, Volume 116, 1992 1206 Figure 9 Indirect double label immunofluorescence of NIH-3T3 cells transfected with tagless pDP ON 48 h after glycerol shock . a and c were stained with DPI-2 .17, to identify mutant DP and b and d were stained with a vimentin polyclonal antibody. In any one population of transiently transfected fibroblasts, a range of phenotypes that appeared to be correlated with the level of mutant expression was observed . The transfected cell (a and b) in which mutant DP coaligned with vimentin networks, is representative of cells that appeared to express lower levels ofDPON . The transfected cell (c and d), in which the vimentin network is disrupted and colocalized with mutant-DP staining, is an example of a cell that appeared to express higher levels of DPAN . Bars, 10 /m .

The colocalization of the carboxyl terminus of DP along (Tsukita and Tsukita, 1985), a 140-kD B related pro- IF also indicates that this domain of DP may be involved in tein (Cartaud et al ., 1990), and plectin (Foisner et al ., 1991) linking IF to the desmosome . The nature of this linkage still are present in much smaller amounts than DP in the desmo- remains to be determined . In vitro binding experiments have some. DP IV, desmocalmin, and the 140-kD lamin B related so far been unable to demonstrate a direct interaction be- protein are likely to be restricted to desmosomes and there- tween DP and IF or any other molecule. One possible expla- fore, in light of the 3T3 results, are unlikely to be absolutely nation for these results, is that the conformation of DP was required for a DP-IF association . This does not exclude the altered as a result of the denaturing conditions used to solu- possibility that they play some sort of stabilizing role in the bilize DP during purification, and that this prevented the di- plaque of the desmosome, thus explaining previous in vitro rect binding of DP to IF. However, it is possible that one or binding data . Certain desmosomal molecules, such as plec- more cellular cofactors or linking proteins may be required tin, are not restricted to desmosomes and are also located in to stabilize and/or mediate the association of DP to IF. This mesenchymal tissues . Therefore we cannot rule out that the would explain the colocalization of DP and IF observed in DP-IF interaction observed in fibroblasts is not stabilized or tissue culture cells when no DP-IF interaction has been mediated by such a protein . reported in vitro . However, the colocalization ofthe carboxyl The coalignment ofDRCT along IF may have implications terminus of DP with vimentin observed in transfected NIH- for the normal process ofdesmosome assembly. Although it 3T3 cells makes it unlikely that these factors are other pro- is not clear if intact IF networks are needed for desmosome teins restricted to desmosomes . formation, previous investigators have suggested that DP is Other molecules localized to desmosomes have been translocated to the cell surface in association with IF during demonstrated to bind IF in vitro . However, one of these, DP the formation of desmosomes (Jones and Goldman, 1985 ; IV, is only present in desmosomes of stratified epithelia Pasdar and Nelson, 1988) . In these experiments desmosome (Kapprell et al ., 1988) while others, including desmocalmin assembly was induced in cultured epithelial cells by switch-

Stappenbeck and Green DP COON Terminus-mediated Collapse of /F 1207 ing cells from low to normal calcium, at which time DPs ap- Evans, M. J., and R. C. Scarpulla . 1988 . Both upstream and intron elements arerequired for elevated expression ofrat somatic cytochrome c gene expres- peared to become redistributed from the cytoplasm to the sion in COS-1 cells. Mol. Cell. Biol. 8:35-41 . cell surface. During the early stages of assembly, DP- Foisner, R., B. Feldman, L. Sander, and G. Wiche. 1991 . Monoclonalantibody containing spots formed a discontinuous linear pattern along mapping of structural and functional plectin epitopes. J. Cell Biol. 112: 397-405 . IF. Ultimately these spots were cleared from the cytoplasm Garrod, D. R., E. P. Parrish, D. L. Mattey, J. E. Marston, H. R. Measures, and DP became restricted to cell-cell interfaces. The and M. J. Vilela. 1990. Desmosomes . In Morphoregulatory Molecules. G. M. Edelman, B. A. Cunningham, and J. P. Thiery, editors. John Wiley colocalization ofthe carboxyl terminus of DP with IF during & Sons Inc., New York. 315-339 . transient transfection may be mimicking the process by Gluzman, Y. 1981 . SV-40 transformed simian cells support the replication of which endogenous DP are assembled into desmosomes. One early SV-40 mutants . Cell. 23 :175-182 . Graham, F. L., and E. Van der Eb . 1973 . A new technique for the assay of possible explanation forthe continuous pattern ofDPCT and infectivity of human adenovirus 5 DNA. Virology. 52 :456-467 . DPON along IF is that the increased level of expression satu- Green, K. J., and J. C. R. Jones. 1990. Interaction of intermediate filaments rated sites of IF association. Consistent with this idea is that with thecell surface. In Cellular and Molecular Biology of Intermediate Fila- ments. R. D. Goldman and P. M. Steinert, editors . Plenum Publishing in cases where lower levels of protein appear to be ex- Corp ., New York . 147-174 . pressed, a more discontinuous, dotty pattern can be ob- Green, K. J., D. A. D. Parry, P. M. Steinert, M. L. A. Virata, R. M. Wagner, B. D. Angst, and L. A. Nilles . 1990 . Structure of the human desmoplakins : served (Fig. 9a). Because COS-7 cells form only a small implications for function in the desmosomal plaque . J. Biol. Chem. number of immature desmosomes as determined by EM, it 265 :2603-2612 . has been difficult to determine the effect ofthe expression of Green, K.J., T. S. Stappenbeck, S. Noguchi, R. Oyasu, and L. A. Nilles. 1991. Desmoplakin expression and distribution in cultured rat bladder epithelial any portion of DP on desmosome formation in these cells. cells of varying tumorigenic potential . Exp. Cell Res . 193 :134-143 . We are currently transfecting individual DP domains into Jones, J. C. R., and R. D. Goldman . 1985 . Intermediate filaments and theinitia- cell lines that have a greater number of mature desmosomes tion of desmosome assembly . J. Cell Biol. 101 :506-517 . Kapprell, H-P., P. Cowin, W. W. Franke, and H. Ponsting . 1985 . Biochemical to study incorporation and/or disruption of desmosomes. characterization of desmosomal proteins isolated from bovine muzzle epidermis: amino acid and carbohydrate composition . Eur. J. Cell Biol. 36 :217-229 . We wish to thank Dr. Michael Klymkowsky for providing 91110.2 anti- Kapprell, H-P., K. Owaribe, and W. W. Franke . 1988 . Identification ofa basic body, Dr. Katrina Trevor for providing Troma I antibody, Dr. Pamela Cow- protein of M, 75,000 as an accessory desmosomal plaque protein in in for providing DPI-2.17 antibody, Dr. Elaine Fuchs for providing sub- stratified and complex epithelia . J. Cell Biol. 106 :1679-1691 . stance P antibody, and Dr. Richard Scarpulla for providing pRC4B. We Kelly, D. E. 1966 . Fine structure of desmosomes, hemidesmosomes, and an adepidermal globular layer in developing newt epidermis . J. Cell Biol. would also like to express our appreciation to Liza Virata, M. Kathleen 28 :51-72 . Rundell, Jonathan Jones, Eileen White, and Rex Chisholm for helpful dis- Klymkowsky, M. W. 1981 . Intermediate filaments in 3T3 cells collapse after cussions and Hue Luu for excellent technical assistance. Special thanks go the intracellular injection of a monoclonal anti-intermediate filament anti- to Walter Glogowski for technical expertise inpreparing specimens for EM. body . Nature (Gond.). 291 :249-251 . Klymkowsky, M. W., R. H. Miller, and E. B. Lane. 1983 . Morphology, be- This work was supported in large part by National Institutes of Health havior, and interaction ofcultured epithelial cellsafter antibody-induced dis- (NIH) grant HD24430, and aided by a March of Dimes Basic Research ruption of keratin filament organization . J. Cell Biol. 96:494-509 . grant 1-PY91-0140 and an American Cancer Society Junior Faculty Re- Klymkowsky, M. W., J. B. Bachant, and A. Domingo. 1989 . Functions of in- search Award to K. J. Green. Additional support was provided by grant termediate filaments. Cell Motil. . 14:309-331 . Kozak, M. 1986 . Point mutations define a sequence flanking the AUG codon No. 2432 from the Council for Tobacco Research, USA, Inc . T. Stappen- that modulates translation by eukaryotic . Cell. 44:283-292 . beck was supported in part by NIH training grant T32-GM08061 . Kulesh, D. A., G. Cecena, Y. M. Darmon, M. Vasseur, and R. G. Oshima . 1989 . Posttranslational regulation of keratins: degradation ofmouse and hu- Received for publication 23 September 1991 and in revised form 9 Decem- man keratins 18 and 8. Mot. Cell Biol. 9:1553-1565 . ber 1991 . Lu, X., and E. B. Lane . 1990 . Retrovirus-mediated transgenic keratin expres- sion in cultured fibroblasts: specific domain functions in keratin stabilization and filament formation . Cell. 62 :681-696 . References Mueller, H., and W. W. Franke . 1983 . Biochemicaland immunological charac- terization of desmoplakins I and 11, the major polypeptides of the des- Albers, K., and E. Fuchs. 1987 . The expression of mutant epidermal keratin mosomal plaque . J. Mol. Biol. 163 :647-671 . cDNAs transfected in simple epithelial and squamous cell lines. Munro, S., and H. R. B. Pelham. 1987 . A C-terminal signal prevents secretion J. Cell Biol. 105:791-806 . of luminal ER proteins. Cell. 48 :899-907 . Albers, K., and E. Fuchs. 1989 . Expression of mutant keratincDNAs in epithe- O'Keefe, E. J., H. P. Erickson, and V. Bennett . 1989 . Desmoplakin I and des- lial cells reveals possible mechanisms for initiation and assembly of inter- moplakin 11: purification and characterization . J. Biol Chem. 264:8310-8318. mediate filaments. J. Cell Biol. 108:1477-1493. Parker, B. A., and G. R. Stark. 1979 . Regulationof simian virus 40 transcrip- Angst, B. D., L. A. Nilles, and K. J. Green. 1990 . Desmoplakin 11 expression tion: sensitive analysis ofthe RNA species present early ininfections by virus is not restricted to stratified epithelia . J. Cell Sci . 97:247-257 . or viral DNA . J. Virol. 31 :360-369 . Amn, J., and L. A. Staehelin. 1981 . The structure and function of spot desmo- Pasdar, M., and W. J. Nelson . 1988 . Kinetics of desmosomal assembly in somes. Dermatology. 20:330-339 . Madin-Darby canine kidney epithelial cells: temporal and spatial regulation Aubin, J. E., M. Osborn, W. W. Franke, and K. Weber. 1980 . Intermediate of desmoplakin organization and stabilization upon cell-cell contact . II. filaments of the vimentin type are distributed differently during mitosis. Exp . Morphological analysis. J. Cell Biol. 106 :687-695 . Cell Res. 129:149-165 . Pasdar, M., K. A. Krzeminski, and W. J. Nelson . 1991 . Regulatio n ofdesmo- Blouin,R., H. Kawahara, S. French, and N. Morceau . 1990 . Selective accumu- some assembly in MDCK cells: coordination of membrane core and cyto- lation of IF proteins at a focal juxtanuclear site in COS-1 cells transfected plasmic plaque domain assembly at the plasma membrane. J. Cell Biol. with mouse keratin 18 cDNA. Exp. Cell Res. 187:234-242 . 113 :645-655 . Bradford, M, M. 1976 . A rapid and sensitive method for the quantitation of Pirbazari, M ., and D. E. Kelly. 1985 . Analysi s of desmosomal intramembrane microgram quantities of protein utilizing the principle of protein-dye bind- particle populations and cytoskeletal elements : detergent extraction and ing. Anal . Biochem . 72:254-259 . freeze-fracture. Cell Tissue Res . 241 :341-351 . Cartaud, A., M. A. Ludosky, J. C. Courvalin, and J. Cartaud . 1990 . A protein Rheinwald, J. G., and T. M. O'Connell . 1985 . Intermediate filament proteins antigenically related to nuclear lamin B mediates the association of inter- as distinguishing markers of cell type and differentiated state in cultured hu- mediate filaments with desmosomes. J. Cell Biol. 111:581-588 . man urinary tract epithelia. Ann. NY Acad. Sci. 455 :259-267 . Conway, J. F., and D. A. D. Parry. 1990. Structuralfeatures inthe heptad sub- Schwarz, M. A., K. Owaribe, J. Kamnbeck, and W. W. Franke. 1990. Desmo- structure and longer range repeats of two-stranded a-fibrous proteins. Inn . somes and hemidesmosomes : constitutive molecular components. Annu . J. Biol. Macromol. 12:328-334. Rev . Cell Biol. 6:461-491 . Cowin, P., H-P. Kapprell, and W. W. Franke . 1985. The complement of des- Shyy, T. T., B. B. Asch, and H. L. Asch. 1989 . Concurrent collapse ofkeratin mosomal plaque proteinsindifferentcell types. J. CellBiol. 101 :1442-1454 . filaments, aggregation of , and inhibition of protein synthesis dur- Evans, G. I., G. K. Lewis, G. Ramsey, and J. M. Bishop . 1985 . Isolation of ing heat shock response in mammary epithelial cells . J. Cell Biol. 108: monoclonal antibodies specific for human c-myc proto-oncogene product . 997-1008 . Mot. Cell. Biol. 5:3610-3616. Steinberg, M. S., H. Shida, G. J. Guidice, M. Shida, N. H. Patel, and O. W.

The Journal of Cell Biology, Volume 116, 1992 1208 Blaschuk . 1987. On the molecular organization, diversity and functions of Mutant keratin expression in transgenic mice causes marked abnormalities desmosomal proteins. In Junctional Complexes ofEpithelial Cells. G. Bock resembling a human genetic skin disease . Cell. 64 :365-380 . and S. Clark, editors . John Wiley & Sons Inc., New York. 3-43 . Virata, M. L. A., R. M. Wagner, D. A. D. Parry, and K. J. Green. 1992. Mo- Steinert, P. M., and D. R. Roop. 1988. Molecular and cellular biology of inter- lecular structure of the human desmoplakin I and II amino terminus . Proc . mediate filaments. Ann. Rev. Biochem. 57:593-625 . Natl. Acad. Sci . USA . 89 :544-548 . Tanaka, T., D. A. D. Parry, V. Klaus-Kovtun, P. M. Steinert, and J. R. Stan- White, E., and R. Cipriani. 1989 . Specific disruption of intermediate filaments ley. 1991 . Comparison of molecularly cloned bullous pemphigoid antigen and the by the 19-kDa product of the adenovirus EIB on- to desmoplakin Iconfirms thattheydefineanew family ofcell adhesion junc- cogene . Proc. Natl. Acad. Sci. USA . 86 :9886-9890 . tion plaque proteins. J. Biol. Chem. 266:12555-12559 . White, E., and R. Cipriani. 1990. Role of adenovirus EIB proteins in transfor- Trevor, K. T. 1990. Disruption ofkeratin filaments in embryonic epithelial cell mation : altered organization of intermediate filaments in transformed cells types. New Biol. 2:1004-1014. that express the 19-kilodalton protein. Mol. Cell Biol. 10:120-130. Tsukita, S., and S. Tsukita . 1985 . Desmocalmin : a calmodulin-binding high Wiche, G., B. Becker, K. Luber, G. Weitzer, M. J. Castanon, R. Hauptmann, molecular weight protein isolated from desmosomes. J. Cell Biol. 101 : C. Stratowa, and M. Stewart . 1991 . Cloning and sequencing of rat plectin 2070-2080 . indicates a 466-kD polypeptide chain with athree-domain structure basedon Vasser, R., P. A. Coulombe, L. Degenstein, K. Albers, and E. Fuchs. 1991 . a central alpha helical . J. Cell Biol. 114:83-99 .

Stappenbeck and Green DP COOH Terminus-mediated Collapse of IF 1209